Real green or fake green? Impact of green credit policy on corporate ESG performance

China has pledged to peak carbon emissions before 2030 and achieve carbon neutrality by 2060. Green credit has evolved into a key policy instrument to guide resource allocation, mitigate environmental risks, and shape corporate behavior, with the 2012 Green Credit Guidelines marking a pivotal institutional moment. The study asks whether the green credit policy effectively improves corporate ESG performance, particularly among heavily polluting enterprises, and through what mechanisms. It highlights the importance of ESG indicators as comprehensive measures of environmental and social performance. The paper notes gaps in the literature, including limited direct evidence on green credit’s effect on ESG and under-exploration of green innovation as a mechanism. It aims to clarify the dual attributes of green credit (traditional finance vs. environmental regulation), theorize the forcing, incentive, and crowding-out effects, and empirically test the policy’s impact using a DID design on Chinese A-share firms (2009–2019).

The literature frames green credit’s impacts through two attributes: (1) Traditional finance attribute—affecting external financing conditions of polluting firms, producing a forcing effect (inducing environmental and social responsibility due to tighter financing) and a crowding-out effect (higher costs diverting resources away from environmental governance and green innovation). (2) Environmental regulation attribute—policy-guiding functions that incentivize better internal controls, greener production, and transparent disclosure, potentially improving ESG outcomes. The net effect depends on the relative strengths of forcing, incentive, and crowding-out effects, leading to Hypothesis 1a (policy enhances ESG) and Hypothesis 1b (policy impedes ESG via crowding out). Regarding mechanisms, green innovation generally correlates positively with ESG performance, but the effect of green credit on green innovation is debated: it could stimulate innovation via internalizing externalities, differentiated financing costs, and longer-term funding, or inhibit it through reduced credit access, information asymmetries, and strategic behaviors (greenwashing, low-quality symbolic innovation), potentially creating a green innovation bubble. Thus, Hypothesis 2a (policy stimulates green innovation and improves ESG) and Hypothesis 2b (policy distorts green innovation and hinders ESG) are proposed. Heterogeneity is expected by ownership and size: SOEs may benefit more given better financing and policy tasks, while non-SOEs face credit discrimination; large firms may better absorb costs than SMEs. Hence, Hypothesis 3a (more conducive for SOEs) and Hypothesis 3b (more conducive for large firms) are posited.

Design: A quasi-natural experiment exploiting the 2012 issuance of China’s Green Credit Guidelines. A difference-in-differences (DID) model compares heavily polluting industries (treatment) with non-polluting industries (control) before and after 2012. The main specification includes firm, year, and province fixed effects and a comprehensive set of controls. Sample: Non-financial A-share listed firms in China from 2009–2019. Polluting firms span 16 industries (e.g., thermal power, steel, cement, electrolytic aluminum, coal). Firms with PT/ST/*ST status or abnormal financials (e.g., asset-liability ratio >1) were excluded. Continuous variables winsorized at 1% two-sided. COVID-19 period excluded by ending at 2019. Variables: Dependent variable is annual Huazheng ESG score (1–9 scale, standardized; annualized from quarterly). Core regressor is post×treat, where post=1 from 2012 onward; treat=1 for heavily polluting industries. Controls include age, assets, R&D intensity, asset-liability ratio, ROA, current ratio, book-to-market, Tobin’s Q, net profit margin, board size, state ownership dummy, CEO duality, and ownership concentration. Data sources: ESG from CNRDS (Huazheng); financials from CSMAR. Identification checks: Event-study for parallel trends with pre-2012 baseline, showing no pre-trend differences and significant post-policy divergence. Fixed effects at firm, year, and province levels included. Robustness: (1) Alternative dependent variables—Huazheng E, S, G subscores (log-transformed), and an alternative ESG source from CNRDS—yield consistent negative effects. (2) Restrict sample to 2009–2016 to exclude potential 2016 green finance guidance shock—results remain negative. (3) PSM-DID with 1:1 and 1:2 nearest-neighbor matching passes balance tests (absolute standardized differences mostly <10%, all <20%) and preserves negative DID coefficients. (4) Placebo tests—counterfactual policy start years (t−1, t−2) and permutation tests (random treatment and years, 500 draws) produce coefficients centered around zero and mostly insignificant, supporting the causal interpretation. Mechanism tests: (a) Crowding-out via financing constraints—moderated DID with SA and WW indices as moderators; interaction post×treat×M centered shows significant negative moderation, indicating stronger negative ESG effects for more constrained firms. (b) Green innovation mediation—two-step mediation with lagged green patent measures: total green patents (pa_s), green utility model patents (pa_um), green invention patents (pa_in). First-stage regressions show post×treat reduces total and invention green patents; second-stage regressions show green patents positively predict ESG and attenuate (but do not eliminate) the negative policy effect, indicating mediation, especially via invention patent (quality) channel.

- Baseline DID: The green credit policy significantly reduces ESG scores of polluting firms. The post×treat coefficient is negative across specifications (e.g., −0.286 to −0.306 with firm/year/province fixed effects), significant at 5–10% levels. - Parallel trends satisfied; post-policy effects persist over time. - Robustness checks (E/S/G subscores, alternative ESG source, pre-2016 sample, PSM-DID) consistently show negative impacts. Placebo (counterfactual timing and random treatment) indicates effects are not driven by unobservables. - Crowding-out mechanism: Interaction of post×treat with financing constraints (SA, WW) is significantly negative (e.g., −0.803 at 1% level; −4.112 at 5%), confirming that tighter financing constraints amplify the policy’s adverse ESG effect. - Green innovation channel: Policy reduces green innovation—total green patents and especially green invention patents decline (core interactions negative; some at 10% significance). Green innovation positively relates to ESG, and mediates the policy effect, with the quality (invention patents) channel more influential than quantity (utility models). - Heterogeneity: The negative effect is stronger and significant for non-SOEs; effects for SOEs are negative but not significant, supporting ownership-based heterogeneity (credit discrimination). By size, SMEs show significant negative effects; large firms do not, but the between-group difference is not statistically significant, so size-based heterogeneity is not confirmed.

The study set out to determine whether China’s 2012 Green Credit Guidelines improved ESG performance among polluting firms and through what mechanisms. Empirical evidence shows the policy reduced ESG performance, indicating that the crowding-out effect from tighter financing dominates the intended forcing and incentive effects. Restricted access to credit leads firms—especially non-SOEs facing stronger credit constraints—to reallocate limited resources away from environmental projects and green innovation, thereby lowering ESG outcomes. The mechanism analysis demonstrates that green innovation is a key pathway: declines in the quality and quantity of green patents, particularly invention patents, depress ESG performance. These findings nuance the dual-nature narrative of green credit: without careful design and implementation that mitigate financing frictions and information asymmetries, the policy’s guidance can be overshadowed by resource crowding-out, undermining ESG objectives. The results are highly relevant to policymakers and financial institutions seeking to align credit policy with environmental performance goals while avoiding adverse distributional consequences across ownership types and firm sizes.

The paper contributes by integrating theoretical and empirical analyses of green credit’s dual attributes and mechanisms, directly estimating policy effects on ESG performance via a DID design and uncovering the mediating role of green innovation. Key conclusions: (1) The 2012 Green Credit Guidelines significantly reduced ESG performance of polluting enterprises. (2) The green innovation channel is pivotal—policy implementation coincided with reductions in green innovation, especially in invention patenting, which in turn depressed ESG. (3) Effects are more adverse for non-SOEs, consistent with credit discrimination; size-based differences are less clear. Policy recommendations: (a) Strengthen supervision and evaluation, with fair rewards/penalties tied to verified environmental project execution and substantive green innovation, to limit rent-seeking and greenwashing and reduce crowding-out. (b) Implement differentiated green credit standards tailored to firm characteristics (e.g., ownership, constraints) to reduce credit frictions for deserving polluting firms transitioning toward greener practices, thereby converting dual effects into dual benefits. Future research could refine ESG and green innovation measurements, incorporate post-2019 policy developments, and explore complementary instruments (e.g., green bonds, guarantees) that alleviate financing constraints while maintaining environmental targets.

- External validity: Results pertain to Chinese A-share listed non-financial firms (2009–2019), particularly heavily polluting sectors, and may not generalize to unlisted firms, other countries, or later periods. - Measurement: ESG performance relies on Huazheng ratings (standardized), and green innovation on patent counts (including classification of “green” and patent types), which may not fully capture qualitative ESG/innovation aspects. - Identification: Although DID with rich fixed effects, event-study, PSM-DID, and placebo tests strengthen causal claims, residual confounding from concurrent policies or unobserved shocks cannot be entirely ruled out. - Time horizon: The analysis ends in 2019 and does not capture COVID-19 and subsequent green finance policy refinements, which could alter effects. - Financing constraint proxies (SA, WW) are indirect and may imperfectly measure firm-specific credit frictions.